Multicentre, randomised, controlled, assessor-blinded

1:1 AIRCHILL plus guideline-based care versus guideline-based care without active AIRCHILL respiratory cooling.

Selected adult OHCA

Adults requiring invasive ventilation after OHCA, unconscious after ROSC, with the pivotal phenotype frozen after Phase II. Initial shockable rhythm is the leading first-pivotal candidate.

90-day neurological outcome

Blinded favourable neurological outcome at approximately 90 days, with full mRS distribution and mortality as key secondary outcomes.

Working Phase-III target for a +7 percentage-point absolute benefit at 90% power, two-sided α=0.05 and 5% missing primary endpoint. Final N is re-estimated from AIRCHILL Phase-II data.

Adults with OHCA who achieve ROSC, remain unconscious and require invasive ventilation, meeting the final prospectively defined phenotype.

AIRCHILL respiratory cooling initiated as early as operationally feasible after the airway is secured, then continued through transport and the protocol-defined early post-ROSC period.

Guideline-based ventilation, post-resuscitation care and temperature management without active AIRCHILL respiratory cooling.

Treatment-policy estimand for the difference in 90-day neurological outcome attributable to assignment to AIRCHILL strategy, with prespecified handling of death, withdrawal of life-sustaining therapy, crossover and missing outcome data.

Temperature × humidity × flow × duration

The final respiratory thermal dose is selected from bench, large-animal, human airway-safety and CA-1/CA-2 data. Record patient-side temperature, absolute humidity, minute ventilation, FiO₂, airway pressure and cumulative exposure continuously or at protocol-defined frequency.

Primary life-support function protected

Prespecified tidal-volume / pressure strategy, PEEP, oxygen and CO₂ targets consistent with guideline care. AIRCHILL cannot be continued through clinically significant ventilation compromise.

Standardise downstream management

Fever prevention, hospital temperature management and rewarming must be protocolised or tightly documented to prevent uncontrolled co-intervention bias.

Prevent self-fulfilling outcome bias

Prespecify delayed, multimodal neuroprognostication and withdrawal-of-life-support safeguards consistent with current post-resuscitation practice.

Central 1:1 randomisation using concealed sequence generation. Variable block sizes; blocks not disclosed to sites.

At minimum site/EMS system and the final key phenotype variable. Candidate factors: initial rhythm, witnessed status and time-to-ROSC, but avoid excessive stratification.

Treating clinicians cannot be blinded. 90-day outcome assessors, adjudication committees and central statisticians should remain masked whenever operationally possible.

All crossover and treatment interruption reasons are timestamped and analysed under intention-to-treat; per-protocol and as-treated analyses remain sensitivity analyses.

Early and recurrent

DSMB/Safety Review Committee reviews unblinded safety after predefined early cohorts and then at regular information fractions. Any emergency feasibility stage can use more frequent review.

One formal interim is preferred

Illustrative Phase-III plan: one interim at ~50% information using a conservative alpha-spending boundary for overwhelming efficacy and a non-binding futility rule. Exact boundary chosen by the trial statistician.

Device, ventilation, haemodynamics, mortality

Predefined triggers for unexpected device-related SAE, ventilation compromise, clinically meaningful excess arrhythmia/hypotension, airway injury signal or a sustained mortality imbalance.

Prefer blinded nuisance-parameter update

A blinded update of control-event rate or outcome variance may be used. Unblinded effect-based adaptation should only be used if prospectively designed with type-I-error control.